Antennas for cellular base stations

challenges, trends and constraints

by Jan Hesselbarth, University of Stuttgart
FP7ARTISAN meeting, Belfast, January 30, 2014
outline:
- choice of frequency
- antenna radiator types and characteristics
- macro sector antennas
- antennas for in-buiding and in-cabin systems
- wireless backhaul
- whats next

January 30, 2014

Jan Hesselbarth

<1>

Antennas to provide coverage, throughput, adaptivity

Different requirements for frequency, pattern, adaptivity, size, cost etc.

macro cell

mm-wave mesh
backhaul
Gbit
hotspot

Gbit
hotspot

Gbit
hotspot

micro cell

January 30, 2014

Jan Hesselbarth

<2>

outline:
- choice of frequency
- antenna radiator types and characteristics
- macro sector antennas
- antennas for in-buiding and in-cabin systems
- wireless backhaul
- whats next

January 30, 2014

Jan Hesselbarth

<3>

Choice of frequency
Criteria for mobile cellular:
antenna size, path loss, diffraction, bandwidth, Doppler
low frequency cellular:

high frequency cellular:

- low path loss / large cells

- strong diffraction
- small Doppler shift
- small bandwidth
- large antennas

- high path loss / small cells

- weak diffraction
- large Doppler shift
- larger bandwidth
- smaller antennas

for coverage and mobility

small throughput

for increased throughput

in rather small cells

(450 MHz) & 700 1000 MHz

1700 2600 MHz

Example: NMT network in Skandinavia at 450 MHz: 25 km cell radius,

few users congested in some cities as early as 1983
January 30, 2014

Jan Hesselbarth

<4>

Choice of frequency
Criteria for wireless backhaul:

[ H.J. Liebe, NTIA report 83-137 ]

path loss, bandwidth, atmospheric attenuation, licensing scheme

January 30, 2014

Jan Hesselbarth

<5>

Choice of frequency
Criteria for wireless backhaul:
path loss, bandwidth, atmospheric attenuation, licensing scheme
long-distance hops (> 10 km) with
raher low capacity at 620 GHz
. dish diameter ~ 12 m
medium-distance hops (310 km) with
high capacity (~ 100 Mbps) at 2844 GHz
. dish diameter ~ 0.6 m
shortest-distance hops (< 1 km) with
multi-GBps capacity at 60 GHz
. dish diameter ~ 0.3 m
short-distance hops (< 2 km) with
multi-GBps capacity at 7186 GHz
. dish diameter ~ 0.3 m
January 30, 2014

Jan Hesselbarth

<6>

outline:
- choice of frequency
- antenna radiator types and characteristics
- macro sector antennas
- antennas for in-buiding and in-cabin systems
- wireless backhaul
- whats next

January 30, 2014

Jan Hesselbarth

<7>

Antenna radiator types and characteristics

For most antennas with sectorial pattern (not: omnidirectional ones),
a groundplane provides suppression of backward radiation and is used
for mounting purposes.
Radiators with groundplane:
patch

dipole over ground

backed slot

all similar in gain (59 dBi) and in beamwidth (90140)

January 30, 2014

Jan Hesselbarth

<8>

Antenna radiator types and characteristics

- backed slot antennas: severe problems with bandwidth
no relevant use
- patch antennas: reasonable bandwidth requires very thick dielectric or
stacked patches;
efficiency and cost and weight requirements lead to
mechanically tricky air dielectric
limited use
example: thick air dielectric patch

[ Alcatel - Lucent ]

(bandwidth ~8% @|S11|< 10dB)

January 30, 2014

Jan Hesselbarth

<9>

Antenna radiator types and characteristics

example: air dielectric stacked patch

(3 resonances bandwidth
~19% @|S11|< 10dB)
patch 2
patch 1
feedline
slot & patches

slot in ground

feed

patch 1

patch 2
January 30, 2014

feed
Jan Hesselbarth

[ Huber+Suhner ]

ground

[ J.-F. Zrcher, F.E. Gardiol, Broadband Patch Antennas, Artech 1995 ]

- patch antennas:

< 10 >

Antenna radiator types and characteristics

January 30, 2014

[ Kathrein ]

[ Huber+Suhner ]

- dipole-over-ground: good radiator bandwdith, wideband balun needed,

rather thick, various low-cost 3D technologies possible
(punched sheet metal, circuit board arrangement,
metalized molded plastic), many PIM-critical connections
widest use

Jan Hesselbarth

< 11 >

outline:
- choice of frequency
- antenna radiator types and characteristics
- macro sector antennas
- antennas for in-buiding and in-cabin systems
- wireless backhaul
- whats next

January 30, 2014

Jan Hesselbarth

< 12 >

Macro sector antennas

- a column of (almost-) in-phase radiators
wide horizonthal (azimuth) beam pattern (3dB BW ~ 60 120)
focused vertical (elevation) beam pattern (3dB BW ~ 612)
linear array of /2, /4-over-ground
dipoles; equal-magnitude, equalphase, 0.8 spacing:
maximum acceptable
spacing is 0.80.9
because of sidelobe
level

6 elements: 15.3 dBi,

SLL -13.2 dB,
3dB BW 10.5
10 elements: 17.5 dBi,
SLL -13.3 dB,
3dB BW 6.3

January 30, 2014

Jan Hesselbarth

< 13 >

Macro sector antennas

- single (V) polarization rarely used
- standard is dual (+45/45) polarization
for diversity receive
- broadband antennas cover about 20% bandwidth

[ Jaybeam ]

[ Kathrein ]

[ Huber+Suhner ]
January 30, 2014

Jan Hesselbarth

< 14 >

Macro sector antennas

[ Kathrein ]

- multi-band antennas have separate radiator columns with separate feeds

January 30, 2014

Jan Hesselbarth

< 15 >

Adaptivity for macro sector antennas

- adaptive downtilt adapts the max cell radius and/or cell edge coverage

5
3

[ Kathrein ]

+3
+5

mechanical downtilt
(can be motorized)

electrical (i.e., motorized) downtilt

January 30, 2014

Jan Hesselbarth

< 16 >

Adaptivity for macro sector antennas

- adaptive beamwidth adapts cell sector width (e.g., 3dB BW 35105)
3dB BW 38

3dB BW 108

25%

0%

50% 25%
(equal phase)

100%

0%

using a motorized differential phase shifter and a 90 hybrid

motorized rotation of the complete radiator

column inside the radome antenna box

January 30, 2014

Jan Hesselbarth

[ Andrew / CommScope ]

- adaptive pan (azimuth beam steering by, e.g., 30)

< 17 >

Adaptivity for macro sector antennas

[ Huber+Suhner ]

- beam steering / switching using multi-column antennas

4 columns
January 30, 2014

Jan Hesselbarth

< 18 >

Adaptivity for macro sector antennas

LightRadio the antenna-integrated base-station
- scalable and flexible
- low OPEX:

. small real estate

. reliable (graceful degradation)
. green (passive cooling)

- low CAPEX: . distributed power amp

. no RF cabling
- all SMT

[ Alcatel - Lucent ]

- multi-radio array:

January 30, 2014

Jan Hesselbarth

< 19 >

Adaptivity for macro sector antennas

[ Ericsson ]

- application of adaptivity: priorization of hot-spot areas

before

after priorization: CDF(SINR) @ 50% : +6.5 dB

January 30, 2014

Jan Hesselbarth

< 20 >

Additional aspects: weight, modularity

- weight: arrays and antennas with active electronics quickly become heavy
. metalized plastics instead of metal
. carbon structures for frames and support
. sandwich structures for maximum stability

[ Huber+Suhner ]

- modularity: antennas with active electronics must be modular

(repair / parts replacement without taken down from tower)
January 30, 2014

Jan Hesselbarth

< 21 >

Additional aspects: passive intermodulation PIM

- PIM: in channelized FDD systems, odd-order IM of two transmit signals
can mask a weak receive signal
example: GSM 1900 (US): UL @ 18501910 & DL @ 19301990
TX1 = 1940 MHz, TX2 = 1980 MHz IM2TX1-TX2 = 1900 MHz

given the receive sensitity, acceptable PIM level must be very small
standard test uses two signals of +43 dBm
measured PIM is at -100-120 dBm, that is, up to -160 dBc !!!
typical specified PIM level of a base station antenna is <150 dBc
PIM measurement & calibration equipment is tricky & expensive

January 30, 2014

Jan Hesselbarth

< 22 >

Additional aspects: PIM

Any electrical non-linearity can (Murphys law: will) cause PIM:
. micro-flashes
- loose metal-to-metal or oxide-to-metal joints
avoid cracks in solder joints or cold solder joints
avoid loosened screws, bolts and connectors
clean production avoid metal dust in the device
. non-linear material
- even some polymers produce PIM
PTFE, PE do not
- corroded metal: e.g., CuO is a known semiconductor
completely (!) plated metal with Sn, Ag, Au
- magnetic material, stainless steel, Co, Ni
avoid galvanic finish (and even PCB) with Ni adhesion layer
use special galvanic processes
low-PIM PCB
- loose metallic building roof installations
place antennas at roof edges
January 30, 2014

Jan Hesselbarth

< 23 >

outline:
- choice of frequency
- antenna radiator types and characteristics
- macro sector antennas
- antennas for in-buiding and in-cabin systems
- wireless backhaul
- whats next

January 30, 2014

Jan Hesselbarth

< 24 >

Antennas for in-building and in-cabin systems

Femto-cells
- signalling / overlapping cells (e.g.: indoors versus outdoors) require
dynamically (e.g., switched) optimized coverage using multiple antennas
- antennas must be cheap;
- high efficiency / low PIM are not required

[ Alcatel - Lucent ]
January 30, 2014

Jan Hesselbarth

< 25 >

in-cabin WiFi (also: GSM1800) does

not benefit from directive antennas
several hotspots will guarantee complete coverage

[ SITA OnAir ]

alternative: leaky cable antenna

[ Kontron ] [ Motorola ]

Antennas for in-building and in-cabin systems

[ Qest ] [ Tecom ] [ row44 ]

[ Gore ]

backhaul (aircraft-to-satellite):
Ku band (UL: 13.75-14.5 GHz,
DL: 10.7-12.75 GHz)

January 30, 2014

Jan Hesselbarth

< 26 >

outline:
- choice of frequency
- antenna radiator types and characteristics
- macro sector antennas
- antennas for in-buiding and in-cabin systems
- wireless backhaul
- whats next

January 30, 2014

Jan Hesselbarth

< 27 >

Wireless backhaul
LTE-tower macro cell backhaul is > 500 MBps gross: optical fiber or mm-wave
small cell backhaul can be anything small, including in-band or copper wire
- wired acces (optical, copper) is preferred if existent (of non-existent, it
is often too time-consuming and/or too expensive to be built)
- in-band backhaul is a waste of precious (0.7 GHz 3 GHz) frequency
- microwave (6 GHz 20 GHz) allows long distance (~ 20 km) but needs
large dishes and has problems with datarates > 100 MBps
- Ka-band (28 GHz 44 GHz) is preferred for macro cell but urban areas
may run out of capacity
- 60 GHz (59 GHz 64 GHz) becomes the best solution for dense
deployments of small cells and fast/ non-permanent installations
- E-band (71 GHz 86 GHz) becomes the best and only solution for
carrier-grade wireless backhaul with >> 1 GBps speed
January 30, 2014

Jan Hesselbarth

< 28 >

The parabolic dish

macro cell & small cell backhaul
- useful formulas:
2 (aperture diameter )

farfield distance:

farfield distance

Friis:

D G = aperture _ efficiency

aperture effiiciency:

aperture _ efficiency =

Kraus:

Tai & Pereira:

DG=

DG=

physical area

maximum effective area

physical area

41253

32dB _ beamwidth

deg

36408

32dB _ beamwidth

January 30, 2014

deg
Jan Hesselbarth

< 29 >

The parabolic dish

macro cell & small cell backhaul
- some typical E-band Cassegrain dishes:

datasheet values
gain [dBi] HPBW

maximum
directivity [dBi]
60 GHz / 90 GHz

200
300
450

39.9
43.5
46.6

1.3
0.9
0.6

42.0
45.5
49.0

45.6
49.1
52.6

aperture
efficiency

directivity from HPBW

Kraus Tai & Pereira

60 GHz / 90 GHz [dBi]

62%
63%
58%

27%
28%
25%

43.9
47.1
50.6

[dBi]
43.3
46.5
50.0

[ Elva-1 ]

diameter
[mm]

- narrow beamwidth requires very accurate alignment and structural stability

- radome loss ranges from nothing at a few GHz to 0.7 dB at E-band

January 30, 2014

Jan Hesselbarth

< 30 >

Planar array versus parabolic dish

Can a planar array replace the parabolic dish ?
- con: the parabolic mirror is a 3D structure and looks like an antenna
- pro: the parabolic mirror is dual polarized and has very low loss
A large planar array (32 x 32 or 64 x 64 elements)
- is flat and square and has larger aperture efficiency (> 90%) than a dish
- loss of the array is in the feed network
- most arrays are single (linearly) polarized
1% efficiency (~ 20 dB loss) of a 32x32 patch
array with microstrip feed network at 60 GHz.

[M. Al Henawy, M. Schneider, Planar antenna

arrays at 60 GHz realized on a new thermoplastic
polymer substrate, Proc. EuCAP 2010]
January 30, 2014

Jan Hesselbarth

< 31 >

Planar array versus parabolic dish

Can a planar array replace the parabolic dish ?
~70% efficiency (1.5 dB loss) of a 32x32 openended waveguide array with ridge waveguide
feed network at 60 GHz.
[Huber+Suhner]

A suspended-substrate slot-coupled square patch array with waveguide feed

two feeder trees for two orthogonal linear polarization combines
all advantages

freq-scaled array building block

January 30, 2014

Jan Hesselbarth

< 32 >

Beam steering for mm-wave wireless backhaul

- small-angle electronic beam steering for ease of alignment
alignment at installation is very expensive (mostly labor costs)
automatic re-alignment would allow for reduced structural stability
- wide-angle electronic beam steering for meshed backhaul network
increase of reliabilty and throughput
same cost reductions as small-angle alignment

January 30, 2014

Jan Hesselbarth

< 33 >

Beam steering for mm-wave wireless backhaul

- any possible solution must provide (reasonably) low cost and low loss
- small-angle electronic beam steering:

dishes with switched

focal plane array for
small-angle electronic
beam alignment
based on inexpensive
feed-horn array and
numbers of switches

mirror

switched
feed
horns

feasible but cost is an issue

January 30, 2014

Jan Hesselbarth

< 34 >

Beam steering for mm-wave wireless backhaul

- wide-angle electronic beam steering:
phased arrays are way too expensive
Butler matrix and Rotman lens are way too lossy
low-loss planar TEM Luneburg lens for 1D scan:
30 GHz planar TE mode air/metal Luneburg lens :

???

[ C. Hua et al., IEEE Trans. MTT, vol. 61, no. 1, January 2013, pp. 436-443 ]
January 30, 2014

Jan Hesselbarth

< 35 >

outline:
- choice of frequency
- antenna radiator types and characteristics
- macro sector antennas
- antennas for in-buiding and in-cabin systems
- wireless backhaul
- whats next

January 30, 2014

Jan Hesselbarth

< 36 >

Whats next massive MIMO

unlike in COMP, phase adjustments are used, similar to holography
massive MIMO
full-dimension MIMO
(incl phase correlation)

single cell

cell split

COMP
(no phase correlation)
January 30, 2014

Jan Hesselbarth

< 37 >

Whats next massive MIMO

massive MIMO antenna consequences:
- large number of radiator columns cost & weight becomes more important
- phase synchronization tricky current use of compact & dense panels
a research topic at its beginnings:
- ok for TD systems, but possible at all for FD systems ?
- wide & sparse panels or fully covered cell circumference much (?) better
- can antennas support synchronization ?
- can non-synchronized repeaters reduce path correlations ?

January 30, 2014

Jan Hesselbarth

< 38 >

Whats next connected arrays

1 / 2

problem: cellular covers a 4:1

frequency range, but
it is useless to develop
4:1 transceivers and
4:1 radiators, because
array element spacing
must be about 0.6 0

solution 1: Nortels dual-band

array

front view

side view

[Nortel, US 6,211,841,B1, 2001 ]

2 / 2

solution 2: connected array

January 30, 2014

Jan Hesselbarth

< 39 >

Whats next connected arrays

linear (1D) 3:1 example:

adaptive
impedance
transformation

coupling
capacitance

very strong (mutual)

coupling and allowing for
complex feed impedances
makes possible, e.g., a / 2
resonance spanning the
complete aperture

tuneable / switched impedance

adaptation can also serve to
conjugate-broadband match the
amplifiers
con: very broadband (multi-octave), but
only one frequency (band) at a given time
January 30, 2014

Jan Hesselbarth

< 40 >

Whats next connected arrays

planar (2D) 4:1 example: identical geometrical aperture area over frequency
1 radiator @ f0

2 x 2 array @ 2 f0

4 x 4 array @ 4 f0

common feeds @ f0 , 2 f0 , 4 f0

maximum directivity & maximum beamforming capability at very

different (here: 4:1) frequencies from a given, common aperture
January 30, 2014

Jan Hesselbarth

< 41 >

Whats next GBps-speed mm-wave UE connections

scenario:
mm-wave directed beams using switched-beam hemispherical hotspot

January 30, 2014

Jan Hesselbarth

< 42 >

Whats next GBps-speed mm-wave UE connections

scenario:
mm-wave directed beams using switched-beam hemispherical hotspot
e.g., 1000 beams of 32 dBi :
on the surface of a sphere: sphere 280
using a graded lens: sphere 14

multitude of patch
arrays on a hemispherical surface

January 30, 2014

principle of
Luneburg
lens

modified Luneburg
lens allowing for
planar feed array

Jan Hesselbarth

< 43 >